A website that links itself — the content graph behind crozzit.de

The pages of this website refer to each other: a problem links the services that solve it. A case study links the problems it demonstrates. Each in German and English.

These links were initially maintained by hand — scattered across many individual files. The consequence was predictable: a new case study was online, but for weeks not a single page linked to it. An orphaned page that neither visitors nor Google can find.

The same pattern as with price lists, product data or master data: distributed maintenance, creeping inconsistency.

A central registry describes the content as a graph: all services, problems and case studies with their relationships — which problem needs which service, which case study demonstrates which problem.

All cross-links are derived from it — in both directions and in both languages. A new entry means: register once, name the relationships — all affected pages link automatically.

The rebuild — designing the registry, rewiring ten data files, adding the index page and navigation entry — was done in one afternoon with AI as a tool, including verification in the browser.

What mattered was not the code but the model: which entities exist, how do they relate? The rest is derivation.

Whether website links, product catalogues or responsibilities: as soon as the same relationship is maintained in several places, it drifts apart. The solution is always the same — model relationships once, derive the presentation.

The "Related topics" box at the end of this page was, by the way, generated by exactly this system.